Post

OpenAI are now under a court order to permanently preserve logs of temporary conversations or paid API usage (previously subject to a 30 day retention policy) - a new twist in the now 17 month lawsuit between the New York Times and OpenAI simonwillison.net/2025/Jun/5/o...

simonwillison.netOpenAI slams court order to save all ChatGPT logs, including deleted chatsThis is very worrying. The New York Times v OpenAI lawsuit, now in its 17th month, includes accusations that OpenAI's models can output verbatim copies of New York Times content …

11 reposts56 likes

One of the main value propositions for the service is their data retention policy. This immediately impacts corporate and government customers who need to have strict requirements for how their data is handled for compliance purposes. It eliminates OAI as a competitive offering.

Okay, but that explains why it's really negative for Open AI and its customers. That's not in itself the same thing as it being absurd. Someone might equally say setting this order aside would be absurd from the point of view of the New York Times, couldn't they?

I showed my lawyers a RAG Chatbot based on Irish public law, and they were sorely impressed. Discussion moved to indexing their client files, and I had to point out that even Mistral holds API call records for 30 days. What are the options for local running embedders?

To determine under what conditions it quotes the source material, and how much of it that it adds relative to the user prompt. There's a big difference between the prompt "explain this article [copy-pasted block from the article]" and "explain x" where it spontaneously recites an article.